An empirical method of estimating the last temperature of water-rock interaction has been devised. It is based upon molar Na, K and Ca concentrations in natural waters from temperature environments ranging from 4 to 340°C. The data for most geothermal waters cluster near a straight line when plotted as the function log (NaK) + β log [ √(Ca)Na] vs reciprocal of absolute temperature, where β is either 13 or 43 depending upon whether the water equilibrated above or below 100°C. For most waters tested, the method gives better results than the NaK methods suggested by other workers. The ratio NaK should not be used to estimate temperature if √ (MCa)MNa is greater than 1. The NaK values of such waters generally yield calculated temperatures much higher than the actual temperature at which water interacted with the rock.

A comparison of the composition of boiling hot-spring water with that obtained from a nearby well (170°C) in Yellowstone Park shows that continued water-rock reactions may occur during ascent of water even though that ascent is so rapid that little or no heat is lost to the country rock, i.e. the water cools adiabatically. As a result of such continued reaction, waters which dissolve additional Ca as they ascend from the aquifer to the surface will yield estimated aquifer temperatures that are too low. On the other hand, waters initially having enough Ca to deposit calcium carbonate during ascent may yield estimated aquifer temperatures that are too high if aqueous Na and K are prevented from further reaction with country rock owing to armoring by calcite or silica minerals.

The Na-K-Ca geothermometer is of particular interest to those prospecting for geothermal energy. The method also may be of use in interpreting compositions of fluid inclusions.